Merge tag 'amd-drm-fixes-6.5-2023-06-30-1' of https://gitlab.freedesktop.org/agd5f/linux into drm-next

amd-drm-fixes-6.5-2023-06-30-1:

amdgpu:
- Misc cleanups
- GFX 9.4.3 fixes
- DEBUGFS build fix
- Fix LPDDR5 reporting
- ASPM fixes
- DCN 3.1.4 fixes
- DP MST fixes
- DCN 3.2.x fixes
- Display PSR TCON fixes
- SMU 13.x fixes
- RAS fixes
- Vega12/20 SMU fixes
- PSP flashing cleanup
- GFX9 MCBP fixes
- SR-IOV fixes
- GPUVM clear mappings fix for always valid BOs
- Add FAMS quirk for problematic monitor
- Fix possible UAF
- Better handle monentary temperature fluctuations
- SDMA 4.4.2 fixes
- Fencing fix

Signed-off-by: Dave Airlie <airlied@redhat.com>

From: Alex Deucher <alexander.deucher@amd.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230630175757.8128-1-alexander.deucher@amd.com

Author: airlied

drivers/gpu/drm/amd/amdgpu/amdgpu.h

@@ -286,6 +286,9 @@ extern int amdgpu_user_partt_mode;
 #define AMDGPU_SMARTSHIFT_MAX_BIAS (100)
 #define AMDGPU_SMARTSHIFT_MIN_BIAS (-100)
 
+/* Extra time delay(in ms) to eliminate the influence of temperature momentary fluctuation */
+#define AMDGPU_SWCTF_EXTRA_DELAY		50
+
 struct amdgpu_xcp_mgr;
 struct amdgpu_device;
 struct amdgpu_irq_src;
@@ -1277,9 +1280,10 @@ int emu_soc_asic_init(struct amdgpu_device *adev);
 
 #define amdgpu_inc_vram_lost(adev) atomic_inc(&((adev)->vram_lost_counter));
 
-#define for_each_inst(i, inst_mask)                                            \
-	for (i = ffs(inst_mask) - 1; inst_mask;                                \
-	     inst_mask &= ~(1U << i), i = ffs(inst_mask) - 1)
+#define BIT_MASK_UPPER(i) ((i) >= BITS_PER_LONG ? 0 : ~0UL << (i))
+#define for_each_inst(i, inst_mask)        \
+	for (i = ffs(inst_mask); i-- != 0; \
+	     i = ffs(inst_mask & BIT_MASK_UPPER(i + 1)))
 
 #define MIN(X, Y) ((X) < (Y) ? (X) : (Y))
 

drivers/gpu/drm/amd/amdgpu/amdgpu_atombios.c

@@ -1791,6 +1791,15 @@ const struct attribute_group amdgpu_vbios_version_attr_group = {
 	.attrs = amdgpu_vbios_version_attrs
 };
 
+int amdgpu_atombios_sysfs_init(struct amdgpu_device *adev)
+{
+	if (adev->mode_info.atom_context)
+		return devm_device_add_group(adev->dev,
+					     &amdgpu_vbios_version_attr_group);
+
+	return 0;
+}
+
 /**
  * amdgpu_atombios_fini - free the driver info and callbacks for atombios
  *

drivers/gpu/drm/amd/amdgpu/amdgpu_atombios.h

@@ -217,5 +217,6 @@ int amdgpu_atombios_get_data_table(struct amdgpu_device *adev,
 
 void amdgpu_atombios_fini(struct amdgpu_device *adev);
 int amdgpu_atombios_init(struct amdgpu_device *adev);
+int amdgpu_atombios_sysfs_init(struct amdgpu_device *adev);
 
 #endif

drivers/gpu/drm/amd/amdgpu/amdgpu_atomfirmware.c

@@ -327,10 +327,13 @@ amdgpu_atomfirmware_get_vram_info(struct amdgpu_device *adev,
 					mem_channel_number = igp_info->v11.umachannelnumber;
 					if (!mem_channel_number)
 						mem_channel_number = 1;
-					/* channel width is 64 */
-					if (vram_width)
-						*vram_width = mem_channel_number * 64;
 					mem_type = igp_info->v11.memorytype;
+					if (mem_type == LpDdr5MemType)
+						mem_channel_width = 32;
+					else
+						mem_channel_width = 64;
+					if (vram_width)
+						*vram_width = mem_channel_number * mem_channel_width;
 					if (vram_type)
 						*vram_type = convert_atom_mem_type_to_vram_type(adev, mem_type);
 					break;
@@ -345,10 +348,13 @@ amdgpu_atomfirmware_get_vram_info(struct amdgpu_device *adev,
 					mem_channel_number = igp_info->v21.umachannelnumber;
 					if (!mem_channel_number)
 						mem_channel_number = 1;
-					/* channel width is 64 */
-					if (vram_width)
-						*vram_width = mem_channel_number * 64;
 					mem_type = igp_info->v21.memorytype;
+					if (mem_type == LpDdr5MemType)
+						mem_channel_width = 32;
+					else
+						mem_channel_width = 64;
+					if (vram_width)
+						*vram_width = mem_channel_number * mem_channel_width;
 					if (vram_type)
 						*vram_type = convert_atom_mem_type_to_vram_type(adev, mem_type);
 					break;

drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c

@@ -136,9 +136,6 @@ static int amdgpu_cs_p1_user_fence(struct amdgpu_cs_parser *p,
 	bo = amdgpu_bo_ref(gem_to_amdgpu_bo(gobj));
 	p->uf_entry.priority = 0;
 	p->uf_entry.tv.bo = &bo->tbo;
-	/* One for TTM and two for the CS job */
-	p->uf_entry.tv.num_shared = 3;
-
 	drm_gem_object_put(gobj);
 
 	size = amdgpu_bo_size(bo);
@@ -912,15 +909,19 @@ static int amdgpu_cs_parser_bos(struct amdgpu_cs_parser *p,
 
 	mutex_lock(&p->bo_list->bo_list_mutex);
 
-	/* One for TTM and one for the CS job */
+	/* One for TTM and one for each CS job */
 	amdgpu_bo_list_for_each_entry(e, p->bo_list)
-		e->tv.num_shared = 2;
+		e->tv.num_shared = 1 + p->gang_size;
+	p->uf_entry.tv.num_shared = 1 + p->gang_size;
 
 	amdgpu_bo_list_get_list(p->bo_list, &p->validated);
 
 	INIT_LIST_HEAD(&duplicates);
 	amdgpu_vm_get_pd_bo(&fpriv->vm, &p->validated, &p->vm_pd);
 
+	/* Two for VM updates, one for TTM and one for each CS job */
+	p->vm_pd.tv.num_shared = 3 + p->gang_size;
+
 	if (p->uf_entry.tv.bo && !ttm_to_amdgpu_bo(p->uf_entry.tv.bo)->parent)
 		list_add(&p->uf_entry.tv.head, &p->validated);
 
@@ -1653,15 +1654,15 @@ static int amdgpu_cs_wait_all_fences(struct amdgpu_device *adev,
 			continue;
 
 		r = dma_fence_wait_timeout(fence, true, timeout);
+		if (r > 0 && fence->error)
+			r = fence->error;
+
 		dma_fence_put(fence);
 		if (r < 0)
 			return r;
 
 		if (r == 0)
 			break;
-
-		if (fence->error)
-			return fence->error;
 	}
 
 	memset(wait, 0, sizeof(*wait));

drivers/gpu/drm/amd/amdgpu/amdgpu_device.c

@@ -2552,7 +2552,7 @@ static int amdgpu_device_ip_init(struct amdgpu_device *adev)
 			adev->ip_blocks[i].status.hw = true;
 
 			/* right after GMC hw init, we create CSA */
-			if (amdgpu_mcbp) {
+			if (adev->gfx.mcbp) {
 				r = amdgpu_allocate_static_csa(adev, &adev->virt.csa_obj,
 							       AMDGPU_GEM_DOMAIN_VRAM |
 							       AMDGPU_GEM_DOMAIN_GTT,
@@ -3673,6 +3673,23 @@ static const struct attribute *amdgpu_dev_attributes[] = {
 	NULL
 };
 
+static void amdgpu_device_set_mcbp(struct amdgpu_device *adev)
+{
+	if (amdgpu_mcbp == 1)
+		adev->gfx.mcbp = true;
+
+	if ((adev->ip_versions[GC_HWIP][0] >= IP_VERSION(9, 0, 0)) &&
+	    (adev->ip_versions[GC_HWIP][0] < IP_VERSION(10, 0, 0)) &&
+	    adev->gfx.num_gfx_rings)
+		adev->gfx.mcbp = true;
+
+	if (amdgpu_sriov_vf(adev))
+		adev->gfx.mcbp = true;
+
+	if (adev->gfx.mcbp)
+		DRM_INFO("MCBP is enabled\n");
+}
+
 /**
  * amdgpu_device_init - initialize the driver
  *
@@ -3824,9 +3841,6 @@ int amdgpu_device_init(struct amdgpu_device *adev,
 	DRM_INFO("register mmio base: 0x%08X\n", (uint32_t)adev->rmmio_base);
 	DRM_INFO("register mmio size: %u\n", (unsigned)adev->rmmio_size);
 
-	if (amdgpu_mcbp)
-		DRM_INFO("MCBP is enabled\n");
-
 	/*
 	 * Reset domain needs to be present early, before XGMI hive discovered
 	 * (if any) and intitialized to use reset sem and in_gpu reset flag
@@ -3852,6 +3866,8 @@ int amdgpu_device_init(struct amdgpu_device *adev,
 	if (r)
 		return r;
 
+	amdgpu_device_set_mcbp(adev);
+
 	/* Get rid of things like offb */
 	r = drm_aperture_remove_conflicting_pci_framebuffers(adev->pdev, &amdgpu_kms_driver);
 	if (r)
@@ -4018,6 +4034,11 @@ int amdgpu_device_init(struct amdgpu_device *adev,
 	/* Get a log2 for easy divisions. */
 	adev->mm_stats.log2_max_MBps = ilog2(max(1u, max_MBps));
 
+	r = amdgpu_atombios_sysfs_init(adev);
+	if (r)
+		drm_err(&adev->ddev,
+			"registering atombios sysfs failed (%d).\n", r);
+
 	r = amdgpu_pm_sysfs_init(adev);
 	if (r)
 		DRM_ERROR("registering pm sysfs failed (%d).\n", r);

drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c

@@ -180,7 +180,7 @@ uint amdgpu_dc_feature_mask = 2;
 uint amdgpu_dc_debug_mask;
 uint amdgpu_dc_visual_confirm;
 int amdgpu_async_gfx_ring = 1;
-int amdgpu_mcbp;
+int amdgpu_mcbp = -1;
 int amdgpu_discovery = -1;
 int amdgpu_mes;
 int amdgpu_mes_kiq;
@@ -634,10 +634,10 @@ module_param_named(async_gfx_ring, amdgpu_async_gfx_ring, int, 0444);
 
 /**
  * DOC: mcbp (int)
- * It is used to enable mid command buffer preemption. (0 = disabled (default), 1 = enabled)
+ * It is used to enable mid command buffer preemption. (0 = disabled, 1 = enabled, -1 auto (default))
  */
 MODULE_PARM_DESC(mcbp,
-	"Enable Mid-command buffer preemption (0 = disabled (default), 1 = enabled)");
+	"Enable Mid-command buffer preemption (0 = disabled, 1 = enabled), -1 = auto (default)");
 module_param_named(mcbp, amdgpu_mcbp, int, 0444);
 
 /**
@@ -2899,12 +2899,10 @@ static struct pci_error_handlers amdgpu_pci_err_handler = {
 
 extern const struct attribute_group amdgpu_vram_mgr_attr_group;
 extern const struct attribute_group amdgpu_gtt_mgr_attr_group;
-extern const struct attribute_group amdgpu_vbios_version_attr_group;
 
 static const struct attribute_group *amdgpu_sysfs_groups[] = {
 	&amdgpu_vram_mgr_attr_group,
 	&amdgpu_gtt_mgr_attr_group,
-	&amdgpu_vbios_version_attr_group,
 	NULL,
 };
 

drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h

@@ -434,6 +434,7 @@ struct amdgpu_gfx {
 	uint16_t 			xcc_mask;
 	uint32_t			num_xcc_per_xcp;
 	struct mutex			partition_mutex;
+	bool				mcbp; /* mid command buffer preemption */
 };
 
 struct amdgpu_gfx_ras_reg_entry {

drivers/gpu/drm/amd/amdgpu/amdgpu_jpeg.c

@@ -255,7 +255,8 @@ int amdgpu_jpeg_ras_late_init(struct amdgpu_device *adev, struct ras_common_if *
 
 	if (amdgpu_ras_is_supported(adev, ras_block->block)) {
 		for (i = 0; i < adev->jpeg.num_jpeg_inst; ++i) {
-			if (adev->jpeg.harvest_config & (1 << i))
+			if (adev->jpeg.harvest_config & (1 << i) ||
+			    !adev->jpeg.inst[i].ras_poison_irq.funcs)
 				continue;
 
 			r = amdgpu_irq_get(adev, &adev->jpeg.inst[i].ras_poison_irq, 0);

drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c

@@ -805,7 +805,7 @@ int amdgpu_info_ioctl(struct drm_device *dev, void *data, struct drm_file *filp)
 		dev_info->ids_flags = 0;
 		if (adev->flags & AMD_IS_APU)
 			dev_info->ids_flags |= AMDGPU_IDS_FLAGS_FUSION;
-		if (amdgpu_mcbp)
+		if (adev->gfx.mcbp)
 			dev_info->ids_flags |= AMDGPU_IDS_FLAGS_PREEMPTION;
 		if (amdgpu_is_tmz(adev))
 			dev_info->ids_flags |= AMDGPU_IDS_FLAGS_TMZ;
@@ -1247,7 +1247,7 @@ int amdgpu_driver_open_kms(struct drm_device *dev, struct drm_file *file_priv)
 		goto error_vm;
 	}
 
-	if (amdgpu_mcbp) {
+	if (adev->gfx.mcbp) {
 		uint64_t csa_addr = amdgpu_csa_vaddr(adev) & AMDGPU_GMC_HOLE_MASK;
 
 		r = amdgpu_map_static_csa(adev, &fpriv->vm, adev->virt.csa_obj,